Membrane Emulsification and Solvent Pervaporation Processes for

May 8, 2012 - We discuss the integration of membrane emulsification and pervaporation processes for the continuous production of functional materials,...
0 downloads 0 Views 5MB Size
Download PDF
Article pubs.acs.org/Langmuir

Membrane Emulsification and Solvent Pervaporation Processes for the Continuous Synthesis of Functional Magnetic and Janus Nanobeads Emily P. Chang and T. Alan Hatton* Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States ABSTRACT: We discuss the integration of membrane emulsification and pervaporation processes for the continuous production of functional materials, such as silica-encapsulated magnetite nanoparticle clusters and asymmetric Janus nanoparticles, by the emulsion droplet solvent evaporation method, which has traditionally been performed in small-scale batch systems. An organic solvent containing primary magnetite nanoparticles (∼10 nm) coated with oleic acid was dispersed in a continuous aqueous phase by membrane emulsification, which enabled the consistent production of nanoparticle-laden solvent droplets of well-controlled size with narrow size distributions. The solvent was removed from the emulsion by pervaporation. Prior to complete solvent removal, the nanoparticle packing density within the clusters was a function of the residence time in the pervaporation unit. The final clusters formed, ∼100−300 nm in size, exhibited the same superparamagnetic behavior as the primary nanoparticles, and were stable in aqueous media with a zeta potential of −70 mV at neutral pH. A facile method was used to coat the nanoclusters with a silica shell, providing sites for surface functionalization with a range of organic ligands. The nanoparticles and clusters were analyzed by a variety of techniques, including TGA, DLS, TEM, EDS, and SQUID. The effects of various parameters, such as the membrane dimensions and flow rate through the unit, on the mass transport rates were elucidated through a parametric modeling study. The applicability of the methods to the production of polymeric beads and more complex particles was demonstrated; to create Janus structures, organic polymer solutions were dispersed as droplets in continuous aqueous phases, and the solvent was subsequently evaporated. The Janus particles consisted either of polymeric cores with magnetite nanoparticles clustered as islands on their surfaces, or of two phase-separated polymers, each constituting half of any given polymeric particle.

■

INTRODUCTION Magnetic nanoparticles are known for their applications in data storage technology, environmental and biomolecular separation, magnetic resonance imaging, drug delivery, enzyme immobilization, and immunoassays.1 Their small length scale confers unique properties not seen in the bulk, including superparamagnetism, which is characterized by high saturation magnetization and zero magnetic remanence. The desire to produce functionalized magnetic nanoparticles commercially is manifest in the rise over the past few decades of a handful of companies ranging from start-ups to global chemical firms that sell magnetic fluids or magnetic beads designed to bind target compounds.2 For example, Turbobeads are cobalt nanoparticles coated with layers of graphene, produced using flame aerosol synthesis;3 they characteristically have broad size and shape distributions. Dynabeads, which are porous polymer particles embedded with iron oxide nanoparticles, are monodisperse but are larger than one micrometer in size and are prepared in small-scale batch systems.4 Our interest is in the development of processes that can produce magnetic nanoparticles of uniform shape and size on a large-scale, continuous basis. It is common to coat magnetic nanoparticles with silica, which prevents nanoparticle aggregation in aqueous solution © 2012 American Chemical Society

over a wide pH range, provides a versatile surface for further functionalization to meet various application needs, imparts biological compatibility, and protects the particles from oxidation and other reactions.5,6 In the absence of the protective silica shells, many organic nucleophilic ligands of interest, such as oximates and imidazoles, chelate strongly with iron ions,7 rendering the nanoparticle core chemically unstable and leading to a loss of magnetization. Similarly, redox-active ligands such as iodosobenzoates and hydroperoxides degrade iron oxide through redox reactions.8 Our group has synthesized magnetic particles functionalized with imidazole derivatives for the decomposition of organophosphates.9 We also investigated the bactericidal properties of silica-encapsulated magnetite nanoparticles functionalized with the antiseptic compound poly(hexamethylene biguanide), and showed that the particles were more chemically stable than their counterparts synthesized without the silica shell.10 The silica coating can be attained via a modified Stöber process, in which the silica precursor tetraethyl orthosilicate (TEOS) is hydrolyzed around Received: March 21, 2012 Revised: April 28, 2012 Published: May 8, 2012 9748

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir

Article

a large membrane surface area to ensure high production rates. A modified technique called premix emulsification has been developed recently, where a premade coarse emulsion is forced through the microporous membrane. This method results in a narrow droplet size distribution at low energy costs and high flows of the dispersed phase (up to 1000 times greater production rate than that for straight membrane emulsification).21−23 The emulsification membranes used in this work were composed of Shirasu porous glass (SPG), which is made by the phase separation of calcium aluminoborosilicate glass synthesized from a volcanic ash called Shirasu.24 The membranes have high mechanical strength and can be regenerated after use with a high-temperature acid treatment. Solvent removal from the emulsions can be completed using pervaporation, a membrane separation process in which the liquid feed is placed in contact with one side of the membrane and the permeated product is removed as a vapor from the other side. The vapor can then be condensed or released as desired. The driving force for the mass transport is the chemical potential gradient across the membrane, and can be created by either applying a vacuum or flowing an inert purge gas on the permeate side. Typical polymeric membranes used include those composed of silicone rubber, cellulose acetate, nitrilebutadiene and styrene-butadiene copolymers, and composite membranes where a permselective layer is deposited on a porous support.25 Pervaporation is commonly used to dehydrate organic solvents, to remove organic compounds from aqueous solutions, and to separate anhydrous organic mixtures.26 In our work, it provides for the efficient and controlled solvent evaporation from an emulsion using tunable parameters such as the unit dimensions and temperature.

magnetite seed particles. However, such a core−shell structure significantly decreases the magnetization of the overall particle, which can make magnetic manipulation difficult. It is possible to decrease the fraction of silica in the overall particle either by replacing the Stöber reaction with the addition of the neat silane compound to the magnetite nanoparticles or by using larger particles as seeds. However, the first method results in large aggregates of the primary nanoparticles with no control over the shape and size; a drawback of the second method is that particles large enough (∼50 nm) to be captured by high gradient magnetic separation (HGMS)11,12 do not possess superparamagnetism, which is present only in nanoparticles between about 2 and 20 nm.13 The clustering of magnetic nanoparticles is a promising approach for the synthesis of magnetic cores for core−shell particles that possess a strong magnetic response even with the silica coating. Bai et al. developed a clustering method where microemulsion oil droplets are used as confining templates within which primary nanoparticles are assembled upon evaporation of low-boiling solvents.14 Other groups have used this approach with combinations of polymers and magnetic nanoparticles to create composite materials.15−18 The clusters, even those that are hundreds of nanometers in size, retain the superparamagnetism of the primary magnetic nanoparticles. The clusters themselves may also act as building blocks for new structures, as they are known to form chains in the presence of a magnetic field.18 Previously, our group produced single- or multidomain crystalline superlattices, amorphous spherical aggregates, and toroidal clusters of magnetite.18 These different morphologies were obtained by emulsifying a hexane phase containing monodisperse magnetite nanoparticles with an aqueous phase containing sodium dodecyl sulfate (SDS), followed by evaporation of the hexane at different temperatures. The emulsification was achieved by ultrasonic homogenization, resulting in a broad range of particle sizes; however, as potential building blocks for new materials, particles of uniform size and shape, with no variation in morphology or chemical heterogeneity, are desired. Ademtech manufactures monodisperse magnetic particles of 100−500 nm using ferrofluid emulsions. The technique involves applying shear to a crude emulsion, thereby elongating the large droplets into long cylinders until they fragment into smaller droplets due to Rayleigh instability. Monodisperse fragmentation is restricted to viscosity ratios in the range 0.01−2 between the dispersed and continuous phases.19 Such methods also require highenergy inputs and cannot be applied to shear-sensitive components such as proteins and starches, which may lose functional properties.20 In this study, membrane emulsification, employing a microporous membrane with a highly uniform pore size as the emulsifying element, was used to obtain better control over the particle size. The droplet size is controlled primarily by the pore size of the membrane and not by the degree of turbulence that causes droplet breakup in more conventional approaches. The dispersed phase is pressurized and forced through the membrane, forming droplets at the pore mouths on the surface of the membrane contacted by the aqueous phase. Advantages of this technique, when compared to conventional turbulencebased methods, are its simplicity, potentially lower energy demands, need for less surfactant, and narrow droplet size distributions.21 Disadvantages include a relatively low flux of the dispersed phase through the membrane, which necessitates

■

EXPERIMENTAL SECTION

Materials. Iron(III) chloride hexahydrate (FeCl3·6H2O) (98%) and iron(II) chloride tetrahydrate (FeCl2·4H2O) (99%), oleic acid (OA) (90%), sodium dodecyl sulfate (SDS) (99%), tetraethyl orthosilicate (TEOS) (99%), poly(propylene carbonate) (PPC) (Mw 50 000), and chloroform were purchased from Sigma-Aldrich Chemical Co. Polystyrene (PS) (Mw 125 000−250 000) was purchased from Alfa Aesar. Ammonium hydroxide (28−30 wt % in solution) was purchased from Acros Organics. Hexane, methanol, and ethanol were purchased from Mallinckrodt, EMD Chemicals, and Pharmoco-AAPER, respectively. All chemicals were used as received. All water utilized in the experiments was Milli-Q (Millipore) deionized water. Synthesis. Preparation of Primary Magnetite Nanoparticles (Fe3O4−OA). Magnetite nanoparticles were synthesized by the coprecipitation of iron(II) and iron(III) chlorides. Briefly, 0.60 g (3 mmol) of FeCl2·4H2O and 1.62 g (6 mmol) of FeCl3·6H2O were added to 30 mL of degassed, deionized water and heated to 80 °C under nitrogen protection. After the iron chlorides were dissolved, 15 mL of a 28% ammonium hydroxide solution was added. The resulting black precipitate was stirred vigorously and then kept at 80 °C under nitrogen protection for one hour. The nanoparticles were separated by an electromagnet (magnetic separator model L-1; S.G. Frantz Co. Inc.) and washed twice with deionized water. They were redispersed in 60 mL of a 1% ammonium hydroxide solution, and oleic acid (24 mmol) was added. The mixture was stirred for one hour. Hydrochloric acid (1 M) was added to bring the pH to a slightly acidic level, leaving an oily precipitate. The nanoparticles were redispersed in 60 mL of hexane and precipitated with excess methanol. They were separated with the electromagnet and dried at 80 °C. Nanoparticle Cluster and Polymeric Bead Formation. Emulsification of 2 mL of 3 wt % Fe3O4−OA nanoparticles in hexane with 70 mL of 0.1 wt % SDS in water was carried out by either ultrasonic 9749

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir

Article

homogenization using a Branson Sonifier 450 sonicator for one minute, or by membrane emulsification at pressures ranging from 320 to 800 kPa. The membrane emulsification was performed using an External Pressure Micro Kit from SPG Technology Co., Ltd., with three hydrophilic membranes, each with a different pore size of 0.1, 0.2, or 0.3 μm, and required between two and three hours. Evaporation of the hexane was achieved using either mechanical stirring at ambient conditions for two days or pervaporation using a mini unit from Applied Membrane Technology, Inc., containing microporous polypropylene hollow fibers coated with polysiloxane, at a flow rate of 0.1 mL/min. Similarly, polystyrene (PS) beads were produced by emulsifying 3 mL of 3 wt % polystyrene in chloroform with 100 mL 1 wt % SDS in water using a 0.2 μm pore membrane. The operating pressure was 310 kPa, and emulsification took about three hours. The chloroform was evaporated under magnetic stirring at ambient conditions. Hybrid Bead Formation. Polystyrene beads partially coated with magnetic nanoparticles were created as follows. Polystyrene and primary magnetite nanoparticles were dissolved and dispersed in chloroform to a concentration of 3 wt % and 1 wt %, respectively. An equal volume of hexane was added. 0.6 mL of this oil phase was emulsified with 10 mL 1 wt % SDS by ultrasonic homogenization for one minute. Polystyrene and poly(propylene carbonate) (PPC) Janus beads were also formed using a similar method, where 0.3 mL of an oil phase comprising PS and PPC (3 wt % each) dissolved in chloroform was emulsified with 10 mL 1 wt % SDS by ultrasonic homogenization. The solvents were removed from the emulsions by pervaporation at a flow rate of 0.1 mL/min. Coating of Magnetic Clusters with Silica. The magnetic clusters were concentrated to about 10 mg/mL, and 2.7 mL of the suspension was placed in 12 mL ethanol. 0.3 mL of 28 wt % ammonium hydroxide was added, and the mixture was sonicated for several minutes. Under sonication, 1 mL of a 0.66% (v/v) solution of TEOS in ethanol was added dropwise. The reaction mixture was stirred vigorously overnight, and then centrifuged at 9500 rpm for 5 min to collect the particles. The particles were washed once with ethanol and then redispersed in ethanol with a final concentration of about 7.5 mg/mL. Particle Characterization. Transmission electron microscopy (TEM) imaging was performed on a JEOL-200CX at an accelerating voltage of 120 kV. High-resolution imaging and energy-dispersive Xray spectroscopy (EDS) were performed on a JEOL-2010 at an accelerating voltage of 200 kV. Samples were prepared by placing a few drops of the nanoparticle dispersion on carbon-coated 200 mesh copper grids by Electron Microscopy Sciences. Superconducting quantum interference device (SQUID) measurements were conducted using a Magnetic Property Measurement System model MPMS-5S (Quantum Design) to determine the magnetization of the particles in an applied magnetic field. All SQUID measurements were performed at 300 K over a field range of −10 to 10 kOe on dried samples weighing 1−5 mg. Thermogravimetric analysis (TGA) measurements were performed on a TGA Q50 (TA Instruments). All measurements were made under a constant flow of nitrogen at 100 mL/min. The temperature was ramped from room temperature to 1000 °C at a rate of 20 °C/min. The sample was dried at 80 °C prior to the measurement. Drop shape analysis to measure the interfacial tension between the oil and aqueous phases was performed using a Krüss DSA 10 MK2 Drop Shape Analysis System. The pendant drop method was used, and the oil phase was squeezed through a U-shaped needle of approximately 1.5 mm diameter to form droplets in the aqueous phase. Zeta-potential measurements were performed using a Brookhaven ZetaPALS zeta-potential analyzer (Brookhaven Instruments Corp). The Smoluchowski equation was used to calculate the zeta-potential from the electrophoretic mobility. Reported values are the averages of ten measurements. Dynamic light scattering (DLS) experiments were performed using a Brookhaven BI-200SM light scattering system (Brookhaven Instruments Corp.) at a detection angle of 90°. Samples were measured for 2 min. The non-negative least-squares (NNLS) algorithm was used to determine the particle size distribution and

distribution averages from the DLS correlation functions. Reported values are the averages of five repeated measurements. Modeling Section. The axial variations in the concentration of solvent in the aqueous phase, CA(z), and the droplet size (or the final cluster size, once the solvent had been removed), Rp(z), were determined by assuming that the system could be treated as if in plug flow, with no radial variations in concentration or size, as reflected in the coupled set of equations given below. The axial variation in the continuous phase concentration is affected both by the loss of solvent by permeation through the membrane and by the gain of solvent from the dissolving droplets.

vz

⎛ ∂R p ⎞ ∂CA 2k + M CA + Npρs vz⎜4πR p2 ⎟=0 ∂z ∂z ⎠ R ⎝

(1)

The droplet size change as solvent is lost by diffusional transport from the droplet surface to the bulk aqueous continuous phase is given by

∂R p ∂z

=−

DA 1 [CAsat − CA(z)] ρs vz R p

(2)

In these equations, vz is the fluid velocity, DA is the diffusion coefficient of the solvent through the aqueous phase, kM is the overall mass transfer coefficient through the membrane, R is the inner radius of a fiber, ρs is the solvent density, and CAsat is the saturated concentration of solvent in water. In eq 2, which gives the mass balance around a single droplet, the mass transfer coefficient is based on the assumption that the Sherwood number is 2, which is valid for dilute dispersions in which we can assume a locally static medium. The membrane is composed of a microporous polypropylene substrate coated with a thin layer of nonporous polysiloxane. Thus, the overall mass transfer coefficient kM can be determined by adding the resistances to transport in the porous and nonporous layers.27 1 1 1 = + kM k p(dlm/d i) SM/Ak np(do/d i) where SM/A is the partition coefficient between the polysiloxane layer and the aqueous phase, do and di are the outer and inner diameters of the fiber, respectively, dlm = (do − di)/[ln(do/di)] is the log mean diameter, and kp = (DA/tp)(ε/τ) and knp = Dnp/tnp are the mass transfer coefficients through the porous and nonporous (polysiloxane) layers, respectively.27 Here, Dnp is the diffusion coefficient of the solvent through the polysiloxane layer; tp and tnp are the thicknesses of the porous and nonporous layers, respectively; and ε is the porosity and τ the tortuosity of the porous layer. The number density is Np = [Vo/(Vo + Vw)]/(4/3πR3p0), where Vo and Vw are the volumes of the dispersed (oil) and continuous (water) phases, respectively, and Rp0 is the initial droplet size. Nondimensionalizing eqs 1 and 2, we obtain

∂θ = − θ + αφ(1 − θ) ∂η

(3)

⎧ 1 ⎪− β (1 − θ) φ > φf ∂φ =⎨ φ ∂η ⎪0 φ ≤ φf ⎩

(4)

with the initial conditions (at η = 0) θ = 1 and φ = 1, where θ, φ, and η are the dimensionless concentration in the aqueous phase, dimensionless particle size, and dimensionless axial position, respectively. We have assumed that the feed aqueous phase is saturated with solvent. The nanoparticles within the droplets imply the existence of a finite final size, φf, which depends on the concentration of nanoparticles in the original solvent phase. The dimensionless variables and constants are 9750

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir θ≡

Article

CA CAsat

α=

ϕ≡

Rp

η≡

R p0

DA R 3 Vo 2 Vo + Vw kM R p20

β=

z (Rνz /2kM) 1 CAsat DA R 2 ρs kM R p20

The aqueous-phase concentration and particle size were scaled with the saturation concentration and the initial particle size, respectively. The dimensionless axial position is the ratio of the characteristic transport time in the axial direction to that in the lateral direction. The solvent flux through the membrane is

■

m″ = kMCA = kMCAsatθ

(5)

RESULTS AND DISCUSSION Preparation of Primary Magnetic Nanoparticles. Magnetite nanoparticles stabilized by oleic acid (Fe3O4−OA) and dispersed in hexane were synthesized and are shown in Figure 1. The TEM image indicates that the nanoparticles were polydispersed, ranging from about 5 to 20 nm. SQUID data, measured at 300 K, show that the nanoparticles were superparamagnetic, with the typical features of zero coercivity and zero remanence. The magnetic behavior of a sample of monodisperse nanoparticles can be described by the Langevin function L(α) = M /Ms = (coth α − 1/α)

(6)

with α = (μ0mH)/(kT), and where M is the magnetization and Ms the saturation magnetization of the collection of dry particles, μ0 is the vacuum magnetic permeability (=1 in Gaussian units), m is the magnitude of the magnetic moment of an individual particle, H is the magnetic field strength, k is the Boltzmann constant, and T is the absolute temperature. The particle size polydispersity can be described by the lognormal distribution function f (y ) =

⎛ −(ln y)2 ⎞ ⎟ exp⎜ 2 yσ 2π ⎠ ⎝ 2σ 1

(7)

where y = D/Dv (D is the particle diameter and Dv is the median diameter of the distribution) and σ is the standard 2 2 deviation of ln y. The standard deviation of y is then [eσ (eσ − 1)]1/2. It has been shown that this distribution generally fits experimental results well for fine particle systems.28 Thus, the magnetic behavior of a polydisperse sample is given by M = Ms

∫0

= Ms

∫0

Figure 1. (a) TEM image of Fe3O4−OA nanoparticles. (b) Magnetic behavior of Fe3O4−OA. Data points are fit with a modified Langevin function accounting for the particle polydispersity. (c) TGA data for Fe3O4−OA, indicating oleic acid contributed about 14% of the mass of the total nanoparticle. (d) TGA data for Fe3O4−OA, where the peaks at 240 and 350 °C suggest interpenetration of oleic acid layers on adjacent nanoparticles.

∞

L(α)f (y)dy ∞

(coth α − 1/α)

⎛ −(ln y)2 ⎞ ⎟ dy exp⎜ 2 yσ 2π ⎠ ⎝ 2σ 1

(8)

similar synthesis methods30,31 and falls within the 30−50 emu/ g range that is typically achieved for magnetite nanoparticles formed by coprecipitation.32 It is lower than the saturation magnetization of the bulk material, because reactions between the carboxylic group of the stabilizing surfactant and the iron oxide surface results in the formation of less magnetic or nonmagnetic layers, in which the spins are disordered and not collinear with the magnetic field.28,29,33,34 This also means that the magnetic diameter of the particles is likely smaller than the physical diameter. The particle size measured by dynamic light scattering is about 43 nm and the distribution is monodisperse, suggesting that clusters of 3−4 nanoparticles form when

We can fit this equation to M vs H data to extract values for the saturation magnetization of the nanoparticles, the average particle size, and the standard deviation of the distribution. The results for the Fe3O4−OA nanoparticles are shown in Figure 1b, with Ms = 42.7 emu/g particle, an average particle size of 11.2 nm, and standard deviation of 0.635. The excellent fit (error = 0.4%) justifies the assumption of the model that the system of particles is noninteracting, and thus, the saturation magnetization of the sample is simply the sum of the individual particle moments.29 The saturation magnetization of the particles is comparable to that found in studies that used 9751

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir

Article

SDS concentration. The measurements for the hexane phase in pure water were consistent with values in the literature for the interfacial tension between pure hexane and pure water of ∼50 mN/m.41 As shown in Figure 2, the final cluster size could be controlled by selecting different membrane pore sizes. The final

dispersed in the solvent. The formation of such small clusters is discussed in previous work.35 Thermogravimetric analysis (TGA) of the Fe3O4−OA nanoparticles reveals that the oleic acid shell contributed about 14 wt % to the total mass of the particles (Figure 1c). The two transitions at 240 and 350 °C in the TGA derivative curves (Figure 1d) suggest two desorption processes, which have been discussed in other works with similarly coated nanoparticles.36,37 Zhang et al. show that the oleic acid is chemisorbed to the magnetite surface as a carboxylate.36 Thus, the fatty acid chains are exposed either to the solvent or to the chains attached to other nanoparticles. The interpenetration of the fatty acid coatings between adjacent nanoparticles accounts for the observed desorption behavior that is characteristic of a bilayer. Such interpenetration of surfactant alkyl tails has been analyzed previously.29 Formation of Magnetic Nanoclusters by Membrane Emulsification and Solvent Evaporation. Spherical clusters of magnetite nanoparticles were formed by the membrane emulsification of a hexane phase containing magnetite nanoparticles in an aqueous phase containing the surfactant sodium dodecyl sulfate (SDS), followed by solvent evaporation. The membranes had pore sizes of 0.1, 0.2, or 0.3 μm; and the transmembrane pressures used for those pore sizes were 800, 480, and 320 kPa, respectively. It is known that the critical pressure needed to force the dispersed phase into the continuous phase is Pc =

Figure 2. Final particle size as a function of membrane pore size for membrane emulsification system.

clusters were reasonably uniform in size, with average relative variations in the range 6−8%. The droplets produced by emulsification were about 6.5 times larger than the pore size, within the typical reported range 2−10 for the droplet to pore size ratio.24 On the basis of the composition of the hexane phase, the final clusters should have been about 15.5% the size of the original droplets in diameter, i.e., of diameter similar to that of the pores, which is indeed what we observed. A comparison of the final cluster sizes, and an indication of variations in these sizes, for clusters obtained using membrane emulsification and those using sonication to form the droplets is shown in Figure 3. The average relative variation for clusters obtained by sonication was about 22%, which is 3−4 times

4γow cos θ Dm

(9)

where γow is the interfacial tension of the O/W interface, θ is the oil contact angle with the membrane surface wetted by the continuous phase, and Dm is the pore size. The operating pressures for the 0.2 and 0.3 μm pores, taken to be those at which dispersed-phase droplets were first observed in the continuous phase as the pressure was slowly increased, were roughly 20−30% higher than the critical pressures. The operating pressure for the 0.1 μm pores was proportionally lower than for the other two, because the maximum pressure allowed for the apparatus was 800 kPa. The oil phase flux through the membrane for all pore sizes was approximately 2.5 L/m2 h. The interfacial tensions between the hexane and aqueous phases were 44, 19, and 7 mN/m for SDS concentrations of 0, 0.1, and 0.75 wt %, respectively. For a contact angle of 0, valid if the pore has straight edges,38,39 the critical pressures for a 0.2 μm pore size should be 880, 380, and 140 kPa for the three SDS concentrations, respectively. It was observed that, at 0.75 wt % SDS, the pores of the membrane became wetted with the aqueous phase, resulting in a fast and uncontrollable flow of the dispersed phase through the membrane and thus a loss of control over the particle size. The surface area per molecule of SDS at saturated adsorption has been reported to be 40−52 Å2,40 and thus, a concentration of 0.1 wt % SDS was sufficient to cover monodisperse droplets with an initial size greater than 200 nm in an emulsion with a dispersed-to-continuous ratio of 3:100 (v/v). Low concentrations of nanoparticles within the hexane phase were not observed to have a significant effect on the interfacial properties between the hexane and aqueous phases, as the interfacial tension was not considerably different for nanoparticle concentrations ranging from 0 to 3 wt % for a given

Figure 3. Comparison of magnetic nanoclusters produced by (a) membrane emulsification and (b) ultrasonic homogenization. 9752

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir

Article

as the suspension dried. In contrast, the clusters that were completely dried by pervaporation rather than on a TEM grid were more spherical and densely packed (Figure 5d). The average particle size, measured by DLS, was roughly 200 nm for all samples after one, two, and three passes. This suggests not that all of the solvent was removed during the first pass, but rather, that the nanoparticles in one solvent droplet began to pack into a fairly dense cluster within the first pass; and it was only the nanoparticles on the outside of the sphere that were not tightly packed until all of the solvent evaporated. Clustering results from a combination of interparticle interactions and confinement due to solvent evaporation. The pervaporation model described by eqs 3 and 4 was solved for a unit containing 180 fibers, each of length 18 cm, inner and outer diameters of 240 and 300 μm, respectively, and a polysiloxane layer thickness of about 400 nm. Typical ranges of the porosity and tortuosity for porous supports are 0.4−0.83 and 1−3, respectively.47 The permeability, which is the product of Dnp and SM/A, of hexane in a polysiloxane membrane is known to be 9.4 × 10−7 cm3(STP)cm/(s cm2 cmHg),48 which is equivalent to about 7.8 × 10−5 cm2/s for an ideal gas at 300 K. Calculations of representative values of the two components of the overall mass transfer coefficient show that the resistance in the porous layer is more than 1000 times greater than that in the nonporous layer. The overall mass transfer coefficient is on the order of 10−4−10−3 cm/s, depending on the porosity and tortuosity of the porous layer. Results are shown in Figure 6 for (2 × 10−3) ≤ kM ≤ (10 × 10−3) cm/s. Initial and final cluster sizes of 1290 and 200 nm and a flow rate of 0.1 mL/min were used. The values of the density, saturation concentration in water, and diffusion coefficient in water of hexane are 0.655 g/ cm3, 1.3 × 10−5 g/cm3, and 1 × 10−5 cm2/s, respectively.49,50 To provide a better physical understanding of the results, the axial position in the following figures is scaled by the length of our unit (18 cm), and thus may be interpreted as the number of passes through the unit. The plots show that CA falls to a level slightly below saturation almost immediately (roughly 0.02% down the length of the unit) (Figure 6a) and then decreases slowly until all of the solvent is removed, at which point CA drops to zero (Figure 6b). As expected, the droplet size shrinks more rapidly as the permeability increases (Figure 6c). Based on our experimental results, we conclude that the mass transfer coefficient for our membrane is about 4 × 10−3 cm/s, since we observed all solvent to be removed between two and three passes through the pervaporation unit. This indicates that the polypropylene layer has a relatively high mass transfer coefficient compared to typical porous supports. Figure 6d shows the cumulative solvent transfer rate as a function of axial position, calculated by integrating the flux, eq 5, over the surface area of the fibers. The flux is directly

higher than the variation seen for the clusters obtained by membrane emulsification. The greater polydispersity appears to be due to the multitude of small solvent droplets (with their dispersed nanoparticles) generated by the ultrasonic mixing. The clusters had the same magnetic behavior as the primary magnetite nanoparticles and were colloidally stable in water with a zeta-potential of about −70 mV at neutral pH. The zetapotential of the clusters as a function of pH is plotted in Figure 4. Even under very acidic conditions, the clusters were

Figure 4. Zeta-potential as a function of pH for magnetic nanoclusters formed by the membrane emulsification of Fe3O4−OA nanoparticles dispersed in hexane with a 0.1 wt % aqueous SDS solution.

negatively charged due to their surface coverage by the anionic surfactant SDS, which has been reported to have a pKa of 1.9.42 The zeta-potential dropped sharply at pH 5, at which point the oleic acid, with a pKa of about 5,43 became negatively charged. This indicates that both oleic acid and SDS were present on the particle surfaces. Under basic conditions, the zeta-potential became less negative with increasing pH. It has been shown that the point of zero charge (PZC) for magnetite nanoparticles synthesized by the coprecipitation of Fe(II) and Fe(III) salts occurs at about pH 8.44 We hypothesize that, at higher pH values, the surface charge of the magnetite became sufficiently negative to cause some of the oleic acid to desorb. This is consistent with our observation that the particles became less charged at pH values greater than 8, where the measurement at pH 12 was still more negative than the measurements at very low pHs. Similar instances of such extrema in zeta-potential can be found in the literature for other species.45,46 Solvent Removal by Pervaporation. Solvent removal by pervaporation was studied for samples produced using ultrasonic homogenization. The change in particle morphology as the solvent was removed by pervaporation is shown in Figure 5. Figure 5a shows clusters immediately after emulsification, when the nanoparticles appear to be loosely associated within the droplets; thus, when placed onto a TEM grid, the nanoparticles spread out to varying degrees as the solvent evaporated from the grid, leaving irregularly shaped aggregates

Figure 5. Magnetic clusters after (a) zero, (b) one, (c) two, and (d) three passes through the pervaporation unit. 9753

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir

Article

Figure 6. Concentration of hexane in the aqueous phase (a) initially falls rapidly to a steady level slightly below saturation and (b) slowly decreases as hexane diffuses out of the membrane. (c) The particle size and (d) the cumulative solvent transfer rate are shown as a function of axial position (number of passes through the 18 cm unit) for varying km through the membrane from 2 × 10−3 to 10 × 10−3 cm/s.

Figure 7. (a,b) Dimensionless concentration of hexane in the aqueous phase, (c,d) dimensionless particle size, and (e,f) cumulative solvent transfer rate as a function of axial position for varying fiber radius and flow rate. The value of R ranged from 60 to 240 μm (for a flow rate of 0.1 mL/min), and the flow rates ranged from 0.05 to 0.2 mL/min (for a fiber radius of 120 μm).

proportional to CA, and hence displays the same behavior; it is fairly constant until the solvent is completely removed,

resulting in a linearly increasing cumulative solvent transfer rate with increasing axial position. The maximum rate is equal 9754

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir

Article

to the flow rate of the oil phase that enters the unit, which is about 0.17 mL/h, and is achieved in all systems from which the solvent is completely removed. The system with the largest mass transfer coefficient results in the fastest solvent transfer, despite having the lowest concentration gradient. For the process used in this work, with kM = 4 × 10−3 cm/s and a flow rate through the pervaporation unit of 0.1 mL/min, the model indicates that the solvent flux was about 0.5 L/m2/h. We also explored the effects of using units of varying sizes and of changing the flow rate through the unit. For example, Figure 7a,c,e shows the results for radii 60, 120, and 240 μm, with a kM of 4 × 10−3 cm/s. The length of the unit, number of fibers, and flow rate through the unit were fixed at 18 cm, 180, and 0.1 mL/min, respectively. As expected, the solvent transfer rate is directly proportional to the total surface area of the fibers. It can be increased 2-fold by doubling the radius of the unit and doubling either the number of fibers or radius of each fiber. Therefore, to reduce the footprint of the pervaporation process, two 9-cm-long units could be used in parallel in place of one 18-cm-long unit of the same radius. The simulation also shows that higher average values of CA can be achieved with larger fiber radii, due to the larger distance the solvent molecules must travel to reach the membrane boundary. Results for different flow rates, with a fiber radius of 120 μm and kM fixed at 4 × 10−3 cm/s, are shown in Figure 7b,d,f, which confirms that, for this model, the residence time required to remove a given amount of solvent is the same regardless of the flow rate; the emulsion must simply be passed through the unit more times, or through a longer unit, for faster flow rates. The solvent concentration in the aqueous phase is constant for all flow rates; the solvent transfer rate per length of the unit is also constant for all flow rates, but the maximum cumulative solvent transfer rate increases proportionally with the flow rate since the solvent enters the unit at a faster rate. Finally, the temperature of the system can be adjusted to achieve different solvent transfer rates. The diffusion coefficient of the solvent through the membrane, Dmem, varies with temperature following an Arrhenius-type relation, that is, that ln(Dmem) varies inversely with (−1/T).51−53 Thus, the solvent transfer rate would be greater at higher temperatures. Moreover, the solubility of hexane in water (CAsat) increases with temperature,54 resulting in a larger driving force for solvent transfer through the membrane. Our simulations (plots not shown) indicate that the solvent transfer rate is indeed proportional to CAsat. Ongoing work is being conducted to quantify the collective effect of temperature on the solvent transfer rate in our system. Figure 8 shows the nondimensional length at which all of the solvent is removed as a function of the two dimensionless constants α and β. This length increases with increasing α and decreases with increasing β. These trends are not surprising given that α contains the fraction of the emulsion that is solvent, and more time would be needed to remove a greater amount of solvent; β contains the saturated concentration of the solvent in the aqueous phase, and higher values of CAsat would reduce the length required due to faster solvent transfer rates. Note that the length scale, as defined in the Modeling section, is about 0.03 cm and that the values of α and β shown for kM = 4 × 10−3 cm/s are 75 and 0.018, respectively. Such a plot can be used to evaluate the combined effects of multiple parameters in the design of a pervaporation unit. Encapsulation of Nanoclusters with Silica. A modified Stöber method55 was used to coat the magnetite clusters with

Figure 8. Dimensionless length of the pervaporation unit at which all solvent is removed, as a function of the dimensionless constants α and β.

silica, where the clusters were used as seeds around which TEOS was hydrolyzed into silica. As shown in Figure 9, a uniform coating was achieved around the clusters. As discussed earlier, the highly basic conditions under which the reaction was performed would decrease the colloidal stability of the clusters; however, adding the TEOS to the reaction mixture under sonication rather than stirring or shaking was observed to prevent the aggregation of clusters that would result in several clusters being coated together. SQUID data indicate that the silica coating decreased the magnetization of the overall particles by roughly 78%. A small part of the loss of magnetization of the overall particle can be attributed to reactions during the encapsulation process that created more nonmagnetic layers near the particle surface; however, the main cause is likely that the silica coating added a significant amount of nonmagnetic mass to the particles. This is consistent with the EDS measurements for elemental composition, which indicate that the coated clusters were approximately 71% silica by mass. The thickness of the shell can be tuned by altering the amount of TEOS added or the reaction time.

Figure 9. Silica-encapsulated clusters produced using a modified Stöber method with magnetic clusters from membrane emulsification as seeds.

The encapsulated clusters were robust and could withstand washing using centrifugation and sonication without breaking apart. Thus, they should be stable in a variety of reaction environments for the attachment of desired ligands. For evidence of the chemical stability of silica-coated magnetite nanoparticles, we refer the reader to works published elsewhere.10 9755

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir

Article

Formation of Polymeric and Hybrid Beads. To demonstrate the applicability of the techniques described above toward the synthesis of a variety of different beads, we also produced purely polystyrene beads, polystyrene beads partially coated with magnetic nanoparticles, and PS/PPC Janus beads, as shown in Figure 10.

Figure 11. Schematic depicting a continuous process for synthesizing monodisperse functional magnetic nanoparticles based on emulsification and solvent evaporation techniques. An oil phase containing primary magnetic nanoparticles is emulsified with an aqueous phase containing surfactant by membrane emulsification. The emulsion is passed through a pervaporation unit to remove the solvent, which is condensed and recycled, forming dense clusters of the primary nanoparticles. Finally, the clusters are pumped into a continuous stirred-tank reactor for encapsulation and other functionalization steps.

Figure 10. (a) Polystyrene beads produced using membrane emulsification, with chloroform as the solvent. (b) Polystyrene beads partially coated with magnetic nanoparticles, formed by removing chloroform and hexane using pervaporation. (c) PS/PPC Janus beads from removing chloroform by pervaporation.

Phase separation occurred in the latter two cases. For the PS/PPC Janus beads, polystyrene and poly(propylene carbonate) are not miscible with each other, although both are soluble in chloroform; hence, once the chloroform was removed, we observed macrophase separation of the two polymers to form the Janus bead. For the beads partially coated with magnetic nanoparticles, chloroform evaporated more quickly than did hexane, and since polystyrene is poorly soluble in hexane, it eventually precipitated to form a bead, while the nanoparticles accumulated on one side of the bead. In such cases with more than one solvent, the ability to control the order in which the solvents are removed by adjusting the solvent concentrations in the purge gas makes the use of pervaporation particularly appealing.

The flux of the oil phase for the production of oil-in-water emulsions typically ranges from 2 to 20 L/m2/h for hydrophilic membranes with an average pore size of 0.2 μm; it has also been suggested to use a pressure between 2 and 10 times the critical pressure in order to achieve a reasonable emulsification time. Such production rates, while slow for many disciplines (e.g., the food industry), are acceptable for the synthesis of “special 'high technology’ products and applications”,21 such as functionalized magnetic nanoparticles. We have shown pervaporation to be a facile and effective method of removing the solvent from the emulsion. The simple model described in this paper can be used to estimate the mass transfer rates of the solvent through the membrane for design purposes. Continuing work includes obtaining fine control over the evaporation rates by fixing the temperature around the pervaporation unit and by using a purge gas with a regulated concentration. As discussed before, a variety of particle morphologies may be obtained under different solvent evaporation conditions,1 and we have shown in this work that membrane emulsification and pervaporation can be used in the production of various polymeric and hybrid beads. It is our intention to extend the concepts discussed for both operations to the development of a continuous process for the synthesis of more complex particles through the emulsification and controlled evaporation of mixed solvent systems.

■

CONCLUSIONS We have introduced the use of membrane emulsification and pervaporation as continuous operations in the production of magnetic clusters of controlled size and shape. To our knowledge, this is the first time that magnetic clusters of primary magnetic nanoparticles were made using the SPG emulsification technique, which has most commonly been used for producing food emulsions or polymeric and silica microspheres for drug delivery and other applications.21,24 Recently, other groups have reported the synthesis of silver nanoparticles56 and polymeric microspheres containing magnetite nanoparticles,57 but in both methods, membrane emulsification was used to create the reaction spaces in which the metal or metal oxide precursors were reacted to form the final product. The method described in this paper does not require any further reactions after the emulsification. Simply, a hexane phase containing primary magnetite nanoparticles (∼10 nm) is dispersed through a membrane of a fixed pore size into an aqueous phase containing a minimal amount of surfactant. Droplets of the hexane phase with narrow size distributions are formed, and dense clusters of magnetite nanoparticles remain after the solvent is removed. The clusters can be easily coated with a layer of silica using the Stöber method, providing them with a protective shell and functionalizable surface. We envision that such operations can be integrated into a continuous process, as depicted in Figure 11.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors were supported by the Defense Threat Reduction Agency and the Singapore-MIT Alliance. 9756

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir

■

Article

(22) Zhou, Q.; Ma, G.; Su, Z. Effect of membrane parameters on the size and uniformity in preparing agarose beads by premix membrane emulsification. J. Membr. Sci. 2009, 326, 694−700. (23) Nazir, A.; Schroën, K.; Boom, R. Premix emulsification: A review. J. Membr. Sci. 2010, 362, 1−11. (24) Nakashima, T.; Shimizu, M.; Kukizaki, M. Particle control of emulsion by membrane emulsification and its applications. Adv. Drug Delivery Rev. 2000, 45, 47−56. (25) Yang, D.; Majumdar, S.; Kovenklioglu, S.; Sirkar, K. K. Hollow fiber contained liquid membrane pervaporation system for the removal of toxic volatile organics from wastewater. J. Membr. Sci. 1995, 103, 195−210. (26) Feng, X.; Huang, R. Y. M. Liquid Separation by Membrane Pervaporation: A Review. Ind. Eng. Chem. Res. 1997, 36, 1048−1066. (27) Abou-Nemeh, I.; Das, A.; Saraf, A.; Sirkar, K. K. A composite hollow fiber membrane-based pervaporation process for separation of VOCs from aqueous surfactant solutions. J. Membr. Sci. 1999, 158, 187−209. (28) Chantrell, R. W.; Popplewell, J.; Charles, S. W. Measurements of Particle Size Distribution Parameters in Ferrofluids. IEEE Trans. Magn. 1978, 14 (5), 975−977. (29) Shen, L.; Laibinis, P. E.; Hatton, T. A. Bilayer Surfactant Stabilized Magnetic Fluids: Synthesis and Interactions at Interfaces. Langmuir 1999, 15, 447−453. (30) Kim, D. K.; Zhang, Y.; Voit, W.; Rao, K. V.; Muhammed, M. Synthesis and characterization of surfactant-coated superparamagnetic monodispersed iron oxide nanoparticles. J. Magn. Magn. Mater. 2001, 225, 30−36. (31) Parvin, K.; Ma, J.; Ly, J. Synthesis and magnetic properties of monodisperse Fe3O4 nanoparticles. J. Appl. Phys. 2004, 95 (11), 7121−7123. (32) Lu, A.; Salabas, E. L.; Schüth, F. Magnetic Nanoparticles: Synthesis, Protection, Functionalization, and Application. Angew. Chem., Int. Ed. 2007, 46, 1222−1244. (33) Yamaura, M.; Camilo, R. L.; Sampaio, L. C.; Macêdo, M. A.; Nakamura, M.; Toma, H. E. Preparation and characterization of (3aminiopropyl) triethoxysilane-coated magnetite nanoparticles. J. Magn. Magn. Mater. 2004, 279, 210−217. (34) Khandhar, A. P.; Ferguson, R. M.; Krishnan, K. M. Monodispersed magnetite nanoparticles optimized for magnetic fluid hyperthermia: Implications in biological systems. J. Appl. Phys. 2011, 109, 07B310. (35) Shen, L.; Stachowiak, A.; Fateen, S. K.; Laibinis, P. E.; Hatton, T. A. Structure of Alkanoic Acid Stabilized Magnetic Fluids. A SmallAngle Neutron and Light Scattering Analysis. Langmuir 2001, 17, 288−299. (36) Zhang, L.; He, R.; Gu, H. Oleic acid coating on the monodisperse magnetite nanoparticles. Appl. Surf. Sci. 2006, 253, 2611−2617. (37) Sahoo, Y.; Pizem, H.; Fried, T.; Golodnitsky, D.; Burstein, L.; Sukenik, C. N.; Markovich, G. Alkyl Phosphonate/Phosphate Coating on Magnetite Nanoparticles: A Comparison with Fatty Acids. Langmuir 2001, 17, 7907−7911. (38) Gijsbertsen-Abrahamse, A. J.; van der Padt, A.; Boom, R. M. Influence of membrane morphology on pore activation in membrane emulsification. J. Membr. Sci. 2003, 217, 141−150. (39) Gijsbertsen-Abrahamse, A. J.; van der Padt, A.; Boom, R. M. Status of cross-flow membrane emulsification and outlook for industrial application. J. Membr. Sci. 2004, 230, 149−159. (40) Kimura, M.; Komaki, M.; Nakajima, T.; Nakahara, H.; Fukuda, K. Influence of Detergent Builders on Surface Properties of Sodium Dodecyl Sulfate Solutions Under Dynamic and Static Conditions. JAOCS 1990, 67 (10), 698−704. (41) Zeppieri, S.; Rodríguez, J.; López de Ramos, A. L. Interfacial Tension of Alkane + Water Systems. J. Chem. Eng. Data 2001, 46, 1086−1088. (42) Dressman, J. B.; Derbin, G. M.; Ismailos, G.; Jarvis, C.; Ozturk, A.; Palsson, B. O.; Wheatley, T. A. Circumvention of pH-dependent

REFERENCES

(1) Woo, E.; Kanagasabai, M. P.; Ahn, I; Lee, C. Synthesis of magnetic/silica nanoparticles with a core of magnetic clusters and their application for the immobilization of His-tagged enzymes. J. Mater. Chem. 2010, 20, 1511−1515. (2) McCoy, M. A New Attraction. C&EN 2009, 24−25. (3) Grass, R. N.; Athanassiou, E. K.; Stark, W. J. Covalently Functionalized Cobalt Nanoparticles as a Platform for Magnetic Separations in Organic Synthesis. Angew. Chem., Int. Ed. 2007, 46, 4909−4912. (4) Ugelstad, J.; Ellingsen, T.; Berge, A.; Helgee, O. B. Magnetic polymer particles and process for the preparation thereof; U.S. Patent Application No. 06/571878, filed December 23, 1983. (5) Yang, D.; Hu, J.; Fu, S. Controlled Synthesis of Magnetite-Silica Nanocomposites via a Seeded Sol-Gel Approach. J. Phys. Chem. C 2009, 113, 7646−7651. (6) Maurice, V.; Georgelin, T.; Siaugue, J.; Cabuil, V. Synthesis and characterization of functionalized core-shell γFe2O3-SiO2 nanoparticles. J. Magn. Magn. Mater. 2009, 321, 1408−1413. (7) Bromberg, L; Hatton, T. A. Nerve Agent Destruction by Recyclable Catalytic Magnetic Nanoparticles. Ind. Eng. Chem. Res. 2005, 44, 7991−7998. (8) Jolivet, J.; Chanéac, C.; Tronc, E. Iron oxide chemistry. From molecular clusters to extended solid networks. Chem. Commun. 2004, 481−487. (9) Bromberg, L.; Chen, L.; Chang, E. P.; Wang, S.; Hatton, T. A. Reactive Silver and Cobalt Nanoparticles Modified with Fatty Acid Ligands Functionalized by Imidazole Derivatives. Chem. Mater. 2010, 22, 5383−5391. (10) Bromberg, L.; Chang, E. P.; Hatton, T. A.; Concheiro, A.; Magariños, B.; Alvarez-Lorenzo, C. Bactericidal Core-Shell Paramagnetic Nanoparticles Functionalized with Poly(hexamethylene biguanide). Langmuir 2010, 27 (1), 420−429. (11) Moeser, G. D.; Roach, K. A.; Green, W. H.; Hatton, T. A. HighGradient Magnetic Separation of Coated Magnetic Nanoparticles. AIChE J. 2004, 50 (11), 2835−2848. (12) Lim, J. K.; Tilton, R. D.; Eggeman, A.; Majetich, S. A. Design and synthesis of plasmonic magnetic nanoparticles. J. Magn. Magn. Mater. 2007, 311, 78−83. (13) Gusev, A. I.; Rempel, A. A. Nanocrystalline Materials; Cambridge International Science Publishing: Cambridge, U. K., 2004. (14) Bai, F.; Wang, D.; Huo, Z.; Chen, W.; Liu, L.; Liang, X.; Chen, C.; Wang, X.; Peng, Q.; Li, Y. A Versatile Bottom-up Assembly Approach to Colloidal Spheres from Nanocrystals. Angew. Chem., Int. Ed. 2007, 46, 6650−6653. (15) Nguyen, H. H.; Nguyen, H. L.; Nguyen, C.; Ngo, Q. T. Preparation of magnetic nanoparticles embedded in polystyrene microspheres. J. Phys.: Conf. Ser. 2009, 187. (16) Urban, M.; Musyanovych, A.; Landfester, K. Fluorescent Superparamagnetic Polylactide Nanoparticles by Combination of Miniemulsion and Emulsion/Solvent Evaporation Techniques. Macromol. Chem. Phys. 2009, 210, 961−970. (17) Gong, T.; Yang, D.; Hu, J.; Yang, W.; Wang, C.; Lu, J. Q. Preparation of monodispersed hybrid nanospheres with high magnetite content from uniform Fe3O4 clusters. Colloids Surf., A 2009, 339, 232−239. (18) Isojima, T.; Suh, S. K.; Vander Sande, J. B.; Hatton, T. A. Controlled Assembly of Nanoparticle Structures: Spherical and Toroidal Superlattices and Nanoparticle-Coated Polymeric Beads. Langmuir 2009, 25 (14), 8292−8298. (19) Schmitt, V.; Leal-Calderon, F.; Bibette, J. Preparation of Monodisperse Particles and Emulsions by Controlled Shear. Top. Curr. Chem. 2003, 227, 195−215. (20) Charcosset, C.; Limayem, I.; Fessi, H. The membrane emulsification process − a review. J. Chem. Technol. Biotechnol. 2004, 79, 209−218. (21) Joscelyne, S. M.; Tragardh, G. Membrane emulsification − a literature review. J. Membr. Sci. 2000, 169, 107−117. 9757

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758

Langmuir

Article

release from ethylcellulose-coated pellets. J. Controlled Release 1995, 36, 251−260. (43) Raymond, K. W. General Organic and Biological Chemistry; John Wiley and Sons: New York, 2009. (44) Tombácz, E.; Majzik, A.; Horvát, Z. S.; Illés, E. Magnetite in Aqueous Medium: Coating Its Surface and Surface Coated with It. Romanian Report in Physics 2006, 58 (3), 281−286. (45) Herrera-Urbina, R.; Fuerstenau, D. W. The effect of Pb(II) species, pH and dissolved carbonate on the zeta potential at the quartz/aqueous solution interface. Colloids Surf., A 1995, 98, 25−33. (46) Suphantharida, P.; Osseo-Asare, K. Cerium Oxide Slurries in CMP. Electrophoretic Mobility and Adsorption Investigations of Ceria/Silicate Interaction. J. Electrochem. Soc. 2004, 151 (100), G658− G662. (47) Krull, F. F.; Fritzmann, C.; Melin, T. Liquid membranes for gas/ vapor separations. J. Membr. Sci. 2008, 325, 509−519. (48) Kuo, A. C. M. Poly(dimethylsiloxane), In Polymer Data Handbook; Oxford University Press, Inc.: Oxford, U.K., 1999 (49) Perry, R. H.; Green, D. W. Perry’s Chemical Engineers’ Handbook; The McGraw-Hill Companies, Inc.: New York, 1999 (50) Price, W. S.; Söderman, O. Self-Diffusion Coefficients of Some Hydrocarbons in Water: Measurements and Scaling Relations. J. Phys. Chem. A 2000, 104, 5892−5894. (51) Zhao, C.; Li, J.; Jiang, Z.; Chen, C. Measurement of the infinite dilution diffusion coefficients of small molecule solvents in silicone rubber by inverse gas chromatography. Eur. Polym. J. 2006, 42, 615− 624. (52) Zhang, Q. G.; Fan, B. C.; Liu, Q. L.; Zhu, A. M.; Shi, F. F. A novel poly(dimethyl siloxane)/poly(oligosilsesquioxanes) composite membrane for pervaporation desulfurization. J. Membr. Sci. 2011, 366, 335−341. (53) Unnikrishnan, G.; Thomas, S. Sorption and Diffusion of Aliphatic Hydrocarbons into Crosslinked Natural Rubber. J. Polym. Sci., Part B: Polym. Phys. 1997, 35 (5), 725−734. (54) Tsonopoulos, C.; Wilson, G. M. High-Temperature Mutual Solubilities of Hydrocarbons and Water. AIChE J. 1983, 29 (6), 990− 999. (55) Stö ber, W.; Fink, A.; Bohn, E. Controlled Growth of Monodisperse Silica Spheres in the Micron Size Range. J. Colloid Interface Sci. 1968, 26, 62−69. (56) Kakazu, E.; Murakami, T.; Akamatsu, K.; Sugawara, T.; Kikuchi, R.; Nakao, S. Preparation of silver nanoparticles using the SPG membrane emulsification technique. J. Membr. Sci. 2010, 354, 1−5. (57) Yang, J.; Hao, D.; Bi, C.; Su, Z.; Wang, L.; Ma, G. Rapid Synthesis of Uniform Magnetic Microspheres by Combining Premix Membrane Emulsification and in Situ Formation Techniques. Ind. Eng. Chem. Res. 2010, 49, 6047−6053.

9758

dx.doi.org/10.1021/la301196z | Langmuir 2012, 28, 9748−9758